Magnetocaloric pump for microfluidic applications

Info

Publication number: 20060292013
Type: Application
Filed: Jun 2, 2005
Publication Date: Dec 28, 2006
Inventors: Lonnie Love (Knoxville, TN), John Jansen (Knoxville, TX), Timothy McKnight (Greenback, TN)
Application Number: 11/143,367

Abstract

A microfluidic magnetocaloric pump. The magnetic and thermal properties of a ferrofluid such as MnZnFe2O4 nanoparticles in an oil- or water-based medium are allowed to interact with a magnetic field that is partially coincident with a thermal gradient. As the ferrofluid heats, it loses its attraction to the magnetic field and is displaced by cooler fluid. The micropump produces fluid propulsion with no moving mechanical parts while requiring only 35 mW power and operation at a temperature of only 40-80° C.

Description

Description

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

The United States Government has rights in this invention pursuant to Contract No. DE-AC05-00OR22725.between the United States Department of Energy and UT-Battelle, LLC.

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to microfluidic pumps. More particularly, a magnetocaloric pump incorporated into an integrated circuit chip utilizes MnZnFe₂O₄ferrofluid. The micropump requires only 35 mW power and operates at a temperature of only 40-80° C.

2. Description of the Prior Art

Microfluidic and lab-on-a-chip devices usually require some form of fluid pumping mechanism. While a great many pumping methods are known, there are only two basic approaches to pumping fluid at the microfluidic level. These are diaphragm and field-induced pumps. Diaphragm pumps use a small diaphragm on the chip that undergoes deformation to move the fluid. Passive valves or restrictions built into the channels control the direction of fluid flow. The diaphragms are deformed using a wide array of actuation technologies including piezoelectric, electromagnetic, electrostatic, thermopnuematics and shape memory alloys. There are two limitations to such approaches. First, the fluid flow is not continuous but pulsating. Second, the pump deforms a membrane at high frequency. Material fatigue limits the life cycle of the pump.

The second approach to pumping fluids at the microfluidics scale utilizes external electric fields to propel the fluid. While this approach provides smooth flow, it generally requires high voltage which impacts electronics packaging and increases the overall system packaging size.

The present invention is based on exploiting the magnetocaloric effect for pumping fluids at the microfluidic level. Many microfluidic applications require thermal cycling of the chemicals. In addition, there are presently many active research programs exploring the use of nanometer sized magnetic particles in chemical analysis systems. Our approach is based on exploiting both the thermal cycling and the magnetic/thermal behavior of the magnetic nanoparticles to provide a fluid propulsion system for microfluidic applications that requires no moving mechanical parts (high reliability) and can be powered with a low voltage, low current power source. The advantage of the low voltage, low current power source is that it will enable further miniaturization and mobility of future microfluidic systems.

REFERENCES

1. E. L. Resler, Jr. and R. E. Rosensweig, “Regenerative Thermoelectric Power,” in Journal of Engineering for Power/Transactions of the ASME, Vol. 89, pp. 399-406, July 1967.

2. R. V. Upadhyay, K. J. Davies, S. Wells and S. W. Charles, “Preparation-and Characterization of Ultra-fine MnFe₂O₄and Mn_xFe_1-xFe₂O₄Spinel Systems: I. Particles,” in J. of Magnetism and Magnetic Materials Vol. 132, pp. 249-257, 1994.

3. C. Aston, “Biological Warfare Canaries,” in IEEE Spectrum, Vol. 38, No. 10, pp. 35-40, October 2001.

4. D. J. Laser and J. G. Santiago, “A Review of Micropumps,” in J. Micromechanics Microengineering., Vol. 14, pp. R35-R64, 2004.

BRIEF SUMMARY OF THE INVENTION

In a first preferred embodiment, a structure for pumping a ferrofluid comprises a fluid channel for containing a MnZnFe₂O₄ferrofluid; a localized heat source in the form of a joule heater near the fluid channel; and a magnetic source near the fluid channel such that the magnetic field of the magnetic source is partially coincident with the thermal field of the heat source.

In another preferred embodiment, a structure for cooling a heat source comprises a fluid channel for containing a MnZnFe₂O₄ferrofluid; the fluid channel located near the heat source to be cooled; and a magnetic source located near the heat source such that the magnetic field of the magnetic source is partially coincident with the thermal field of the heat source, and the heat source is cooled by flow of the ferrofluid due to interaction of the heat source and magnetic source with the ferrofluid.

In another preferred embodiment, a self-regulating method for cooling a heat source comprises the steps of locating a fluid channel containing a MnZnFe₂O₄ferrofluid near a heat source to be cooled; and locating a magnetic source near the heat source such that the magnetic field of the magnetic source is partially coincident with the thermal field of the heat source, and cooling of the heat source is due to flow of the ferrofluid caused by the interaction of the heat source and the magnetic source with the ferrofluid.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a graph showing the temperature effect on magnetism for various materials.

FIG. 2 illustrates the interaction of a magnetic field and a thermal field with a ferrofluid in a channel.

FIG. 3 is a graph of the temperature variation in magnetization of a magnetite ferrofluid.

FIG. 4 is a graph of the temperature variation in magnetization of a MnZn ferrofluid.

FIG. 5 is an exploded view of magnetocaloric structure in accordance with the invention.

FIG. 6 illustrates the interaction of a heat source, a magnetic source, and a ferrofluid to produce movement of the ferrofluid in a fluid channel.

DETAILED DESCRIPTION OF THE INVENTION

A magnetocaloric pump provides a simple means of pumping fluid using only external thermal and magnetic fields. The principle, which can be traced-back to the early work of Resler and Rosensweig (Ref. 1), is straightforward. Magnetic materials tend to lose their magnetization as the temperature approaches the material's Curie point (FIG. 1). Exposing a column of magnetic fluid to a uniform magnetic field that is partially coincident with a temperature gradient produces a pressure gradient in the magnetic fluid (FIG. 2). As the fluid heats up, it loses its attraction to the magnetic field and is displaced by cooler fluid.

Magnetocaloric pumping is based on the temperature dependence of magnetic materials. Magnetization of materials is based upon alignment of-the magnetic moments of individual atoms. Ferromagnetic materials contain domains in the material where there is consistent alignment of the magnetic moments. When exposed to an external magnetic field, these domains grow, increasing the net magnetization (M) of the material. However, sufficient thermal agitation disrupts the alignment of magnetic moments, reducing the net magnetization.

FIG. 1 shows the degradation in magnetization as a function of temperature. The temperature at which all net magnetization is lost is called the Curie temperature, T_c. This temperature is a function of the atomic density, m, material magnetic dipole moment, μ_m, permeability of air, μ_o, and Boltzmann's constant, k. Table 1 lists the saturation magnetization and Curie temperature of a variety of known magnetic materials. $T_{c} = \frac{m μ_{0} μ_{m}^{2}}{3 k}$ $M = m μ_{m}$

TABLE 1 Curie temperature and saturation magnetization of ferromagnetic solids. Substance Curie Temp (° C.) μ₀M_s(T) Dysprosium −185 3.67 Gadolinium 19 2.59 Nickel 358 0.64 Magnetite 585 0.56 Iron 770 2.18 Cobalt 1120 1.82

Ferrofluids are oil- or water-based liquids that are loaded with nanometer-sized ferrous particles. The particle size plays a significant role in the fluid's behavior. Unlike magnetorheological (MR) fluids, ferrofluids remain in suspension, even during extended exposure to magnetic fields. Thermal agitation in the fluid, even at room temperature, exceeds the gravitational potential of the particles. Ferrofluids are based on magnetic nanoparticles that are so small they contain only a single magnetic domain. While temperature may not disrupt the magnetic domain in the particles, it can disrupt any alignment between adjacent particles. Thus, ferrofluids experience a similar demagnetization as a function of temperature.

As previously mentioned, magnetocaloric energy conversion was first described by Resler and Rosensweig (Ref. 1). The principle of operation, shown in FIG. 2, is relatively simple. In FIG. 2, a ferrofluid 11 in a channel is attracted to a magnetic field 13. This can be from a coil or permanent magnet. The ferrofluid 11 encounters a thermal field 15, causing the temperature of the ferrofluid and nanoparticles to increase. As the temperature of the ferrofluid approaches the Curie temperature, the material's local magnetization decreases (see FIG. 1). The cooler ferrofluid 11, which is attracted to the magnetic field 13, displaces the exiting warmer fluid 17. The result is fluid flow achieved using only the external magnetic and thermal fields 13 and 15, respectively.

The ideal magnetic nanoparticle would have a Curie temperature close to the maximum system operating temperature. Most commercial grade ferrofluids are based on magnetite particles. While initially attractive from a fabrication perspective (cost), the Curie temperature of magnetite particles is far above the expected operating temperature of many conventional fluidic systems. Most fluids would boil before the temperature would significantly reduce the material's magnetization.

Various ferrite compositions have been synthesized as nanoparticles. In particular, manganese zinc ferrite particles (Mn_xFe_1-xFe₂O₄ferrites) have been reported to have Curie temperatures between 75 and 325° C., and to have as-synthesized particle sizes between 6 and 20 nm. The physical properties of this doped magnetite fit well with our requirements for ferrofluid particles. FIGS. 3 and 4 display the variation in magnetization, for Magnetite and MnZn ferrofluids respectively, as the temperature varied from 25° C. to 85° C. in 10° C. increments. It is clear that the MnZn ferrofluids exhibit a much larger variation in magnetization for the given temperature range. As described previously, this temperature sensitive variation in magnetization generates pressure gradients in the fluid. Subsequently, we should see significant improvements in fluid flow by simply modifying the nanoparticles used in the ferrofluid.

Experimental Investigation

In a comparison of magnetocaloric energy conversion, oil and water based ferrofluids are compared using the same series of particles, Mn_0.5Zn_0.5Fe₂O₄, which has a Curie temperature of approximately 150° C. It is advantageous to design a ferrofluid with a low viscosity, high specific heat, large pyromagnetic coefficient and Curie temperature close to the maximum cycle temperature. The wide array of potential applications may require either an oil-based or water-based carrier fluid.

An experiment was done to establish the impact the ferrofluid had on the measured flow rate. A 2 mm diameter glass tube with a 40 mm long column of ferrofluid was tested. The heat source was a coil of nichrome wire. The coil was wrapped around the glass tube, was approximately 15 mm long, and had a 4.4Ω resistance. We compared three different series of ferrofluids: magnetite in oil, Mn_0.5Zn_0.5Fe₂O₄in oil and Mn_0.5Zn_0.5Fe₂O₄in water. The experiments were controlled so that the only variable that changed was the ferrofluid. Thermocouples measured the fluid temperature at the heat source and at the end of the fluid column. We used the same heating element with the same excitation (2.6 V, 0.56 A) for each experiment. The average field strength across the column of fluid was 2.7e5 A/m. We used a digital video camera to record the experiment. We estimated the fluid velocity, via the digital video output, by measuring the displacement, observed with a ruler, and time, recorded on the time stamp on the video. The results of these experiments are tabulated in Table 2.

TABLE 2 Magnetocaloric pump comparison. V_x Specific Viscosity Ms T_low T_hi (FEA) V_x Material Grav. (mPa-s) (mT) (° C.) (° C.) (mm/s) (mm/s) Fe₃O₄in oil 1.40 375 35 74 86 0.23 0.17 Mn_0.5Zn_0.5Fe₂O₄in oil 1.52 380 25 48 61 1.60 1.59 Mn_0.5Zn_0.5Fe₂O₄in water 1.37 83 11 46 59 1.57 2.10

In Table 2, the last two columns represent the flow estimate from the finite element analysis and the measured flow. The first observation is that we achieved approximately an order of magnitude increase in flow, at lower temperatures, than was previously possible with conventional magnetite-based ferrofluids. For the estimate, we are assuming the field is constant across the diameter of the tube. While we know the amount of power generated by the heating element, there is some uncertainty in how much of this heat is transferred to the fluid. To overcome this problem, we used thermocouples at the inlet and outlet to measure the temperature gradient.

We explored the possibility of improving our velocity estimates with finite element analysis. We used the same constitutive relationships described above for the magnetic, thermal and fluid dynamic models, but use finite element analysis to refine the spatial variations in the constitutive relationships. This problem is complicated somewhat in that there is significant coupling between each of the three domains (thermal, magnetic, and fluid dynamics). We have already discussed how the magnetic and thermal fields produce pressure gradients in the fluid. However, fluid flow in the channel impacts the thermal distribution. In addition, the fluid has a temperature dependent magnetic susceptibility that impacts the magnetic field. To include each of these phenomena, we used a multiphysics finite element package, Femlab. The analysis for our water based Mn_0.5Zn_0.5Fe₂O₄ferrofluid was done under the same conditions as described previously. The results are tabulated in Table 2 as well. The finite element analysis (FEA) predicts velocities anywhere from 5% to 53% over the actual flow. It is possible to refine the models more to include variations in viscosity due to magnetic and thermal variations.

Microfluidic Application

For now, our motivation is to develop models that provide reasonable estimates of fluid flow and can aid in the design of systems that exploit the magnetocaloric effect. As an example, we consider the design of a microfluidic pump for “lab-on-a-chip” (LOC) applications.

A popular method for controlling the flow of fluids in LOC systems is based on deforming diaphragms driven by piezoelectric, thermopneumatic, electrostatic, electromagnetic, and shape memory actuation. The advantage of this approach to micropump design is the independence of the fluid medium. However, the deformable diaphragm approach does require deflecting a material at a high frequency. This has two drawbacks. First, the fluid flow is pulsating, not continuous. Second, the fatigue life can be relatively short. For example, one reported shape memory alloy (SMA) pump has a fatigue life of approximately 4×10⁷cycles. The operating frequency is 100 Hz, leading to approximately 111 hours of operation before probable failure. This limitation has led to more recent work that focuses on field-induced pumps that require no moving mechanical parts.

Another LOC system utilizes an electroosmotic (ion drag) pump based on high differential voltage between inlet and outlet channels. However, while the electroosmotic approach is attractive based upon a lack of moving mechanical parts, this approach requires high voltage levels (˜1000 V), complicating miniaturization of the power electronics.

Field induced flow based on ferrofluids has several advantages in an LOC configuration. First, many DNA amplification and chemical processes require thermal cycling (heating and cooling) of the fluid that could also serve as the source of the thermal gradient for the field induced pumping. Second, magnetic sensing and detection systems rely on magnetic sensors and microbeads to detect the presence and concentration of bioagent DNA. It is possible to use the same nanoparticles for both the magnetic sensing and field induced pumping. Third, this methodology requires no moving mechanical parts (increasing reliability) and requires only the addition of an external magnetic field. There is no need for a high voltage source.

In a test of our invention, a chip structure with a 50 micron (approximately the width of a single hair) by 10 micron channel loaded with a 15 mm column of oil based Mn—Zn ferrofluid was utilized. The average flow rate, when the thermoelectric heater was energized, was 2 microns/sec. The finite element model provided a close match (1.4 microns/sec) to the actual fluid velocity.

In the design process, we considered two methods for supplying the heat: an external thermoelectric heater and an integrated joule heating element. The external thermoelectric heater provided the ability to modify existing chips with a commercial heating element. However, our finite element analysis and experimentation showed that this approach lacks in efficiency. The heat must first pass through the glass before heating the fluid. This approach to heating lacks the ability to focus the heat directly on the fluid. In contrast the finite element analysis shows that, by locating the integrated joule heating element in near contact with the fluid channel, we could better control the heat distribution on the column of ferrofluid.

We should expect approximately an order of magnitude increase in flow rate if we use water based Mn—Zn ferrofluids and move the heating element directly adjacent to the column of fluid. In a preferred embodiment of our invention, shown in FIGS. 5 and 6. we used photolithographic techniques and a metal liftoff process to pattern gold heating coils 31 directly onto a glass substrate adjacent the fluid channel 35. In the example of FIG. 5, the fluid channel 35 was formed in a glass layer 37 that was bonded to the glass substrate 33 to form the fluid channel 35. The magnet 39 was placed opposite the fluid channel 35 from the heating coils 31. In addition to being placed opposite the heating coils 31, the magnet 39 was located such that its magnetic field is at least partially coincident with the thermal field from the heating coils 31. This is shown in FIGS. 2, 5 and 6. FIG. 6 illustrates the use of a passivation layer 41 between the fluid channel 35 and heating coils 31.

The integrated joule heating element 31 can be etched in glass and the microfluidic channel 35 etched in the adjoining glass surface 37. We compared three materials for the channel 35. The first, glass, works well with oil and water based ferrofluids, but was not flexible and difficult to clean. The second material, SU-8, worked well with water and oil, permanently attaches to the glass, but was likewise difficult to clean. The final material, PDMS, worked well with water based ferrofluids, was easy to clean and reuse, but proved difficult to work with oil based fluids.

There are three advantages to this approach. First, the heat is directly focused on the fluid, reducing the amount of power required for heating. In our chip structure using MnZnFe₂O₄ferrofluid, the heat source need only heat the ferrofluid to 40°-80° C. for the ferrofluid to lose magnetization. Our required power for heating drops from 1.5 W to 35 mW.

This reduction in the power requirement leads to the second advantage of this approach. A 3V or 9V battery, similar to those used for hand held calculators, provides enough current to heat the fluid to our target levels.

In order to avoid generation of gaseous species from electrolysis of the water-based ferrofluid, a passivation layer 41 comprised of 100 nm of silicon dioxide was deposited upon the gold metallization 31 using plasma-enhanced chemical vapor deposition. The passivation layer 41 could also be: LPCVD (low pressure chemical vapor deposited), PECVD (plasma enhanced chemical vapor deposited), SiO₂/Si_xN_y, evaporated passivation, sputtered, a passive oxide layer that forms on certain metals, electrochemical oxidized, etc. It could also be spun on glass or photoresist, including Su-8.

The third advantage is that the heating is localized. All of the previous analysis shows considerable agreement between the finite element model of the magnetocaloric energy pump and the experimental results.

It is possible to have multiple stages of partially coincident magnetic and thermal zones. The finite element analysis aides in quantifying the spacing between these zones. This spacing provides an effective fluid column length for each stage. Since the flow is laminar, the flow rate is inversely proportional to the column length. By having multiple stages, it is possible to effectively reduce the fluid column length for each stage, significantly increasing the fluid flow rate.

The nanoparticles described herein have the potential for a secondary use in sensing and detection. Researchers are presently exploring the ability to coat these magnetic nanoparticles with materials that will selectively attach the nanoparticles to specific organisms and cells, thereby providing the potential for magnetic detection of single molecules and cells. Subsequently, the same magnetic nanoparticles used for detection are used for fluid pumping, a dual use for the ferrofluid.

Cooling Embodiment

The basic principle for cooling is exactly the same as the pump except now the fluid acts as a convective heat exchanger taking heat away from a hot component, represented by the thermal field in FIG. 2 or the heat source in FIG. 6. As in the pump embodiment, cool fluid is attracted to the magnet. The hot component heats the fluid which subsequently loses it's attraction to the magnetic field and is displaced by cooler fluid. Subsequently, the pump is now acting as a heat pump removing heat from the hot component. The advantage of such an approach is 1) no moving mechanical parts and 2) self regulating control. As the hot component increases in temperature, the fluid flow will increase naturally, increasing the heat removal rate.

All electromagnetic actuators have coincident magnetic and thermal fields. The maximum operating torque of electric actuators is limited due to thermal constraints on the coils. Using this invention, it is possible to cool an electric motor using ferrofluids. The advantage to such an approach is two-fold. First, there is no need for an external pump because the internal magnetic and thermal fields provide the energy for fluid propulsion. Second, the system is self regulating. As the motor temperature increases, the fluid flow increases. Such applications require the use of oil based ferrofluids for electrical insulation.

Claims

1. A structure for use with a ferrofluid comprising:

a fluid channel for containing a ferrofluid;

a heat source near said fluid channel; and

a magnetic source near said fluid channel such that the magnetic field of said magnetic source is partially coincident with the thermal field of said heat source.

2. The structure of claim 1 wherein said ferrofluid is MnZnFe2O4 nanoparticles in a liquid medium.

3. The structure of claim 2 wherein said liquid medium is oil.

4. The structure of claim 2 wherein said liquid medium is water.

5. The structure of claim 1 wherein said heat source is a joule heater for generating flow of the ferrofluid within the fluid channel due to the interaction of said heat source and said magnetic source with the ferrofluid in said fluid channel.

6. The structure of claim 1 wherein said heat source is cooled by flow of the ferrofluid, and wherein said ferrofluid flow is caused by the interaction of said heat source, said magnetic source with the ferrofluid.

7. The structure of claim 6 wherein said heat source is an electric generator.

8. The structure of claim 6 wherein said heat source is an electric motor.

9. The structure of claim 8 wherein said electric motor is a brushless dc motor.

10. A self-regulating method for cooling a heat source comprising the steps of:

locating a fluid channel containing a ferrofluid near a heat source; and

locating a magnetic source near said fluid channel such that the magnetic field of said magnetic source is partially coincident with the thermal field of said heat source, and cooling of the heat source is due to flow of the ferrofluid caused by the interaction of said heat source and said magnetic source with the ferrofluid.